
# %pip install python-dotenv pandas numpy nbformat scipy scikit-learn statsmodels sqlalchemy plotly


from dotenv import load_dotenv

load_dotenv()

from sqlalchemy import create_engine
import sqlite3
from math import sqrt
import pandas as pd
import numpy as np

from scipy.stats import pointbiserialr, f_oneway, chi2_contingency, spearmanr
from statsmodels.stats.multicomp import pairwise_tukeyhsd
from sklearn.metrics import r2_score

import plotly
import plotly.express as px
import plotly.io as pio
import plotly.graph_objects as go
import plotly.subplots as sp

pio.renderers.default = "notebook"
plotly.offline.init_notebook_mode()


from utils.functions import (
    clean_gender_data,
    clean_age_data,
    clean_to_numerical,
    database_filtering,
    create_bar_plot,
)


cnx = sqlite3.connect("./datasets/mental_health.sqlite")

df = pd.read_sql_query("SELECT name FROM sqlite_master WHERE type='table'", cnx)

print(df)

       name
0    Answer
1  Question
2    Survey


answer = pd.read_sql_query("SELECT * FROM Answer", cnx)
question = pd.read_sql_query("SELECT * FROM Question", cnx)
survey = pd.read_sql_query("SELECT * FROM Survey", cnx)

display(answer.head())
display(question.head())
display(survey.head())


print("Missing values in Answer table:")
print(answer.isnull().sum())

print("\nMissing values in Question table:")
print(question.isnull().sum())

print("\nMissing values in Survey table:")
print(survey.isnull().sum())

print("\nDuplicates in Answer table:")
print(answer.duplicated().sum())

print("\nDuplicates in Question table:")
print(question.duplicated().sum())

print("\nDuplicates in Survey table:")
print(survey.duplicated().sum())

Missing values in Answer table:
AnswerText    0
SurveyID      0
UserID        0
QuestionID    0
dtype: int64

Missing values in Question table:
questiontext    0
questionid      0
dtype: int64

Missing values in Survey table:
SurveyID       0
Description    0
dtype: int64

Duplicates in Answer table:
0

Duplicates in Question table:
0

Duplicates in Survey table:
0


merged_table = pd.merge(
    answer, question, left_on="QuestionID", right_on="questionid", how="outer"
)
merged_table.drop("questionid", axis=1, inplace=True)

merged_table = pd.merge(merged_table, survey, on="SurveyID", how="outer")
merged_table.rename(columns={"questiontext": "QuestionText"}, inplace=True)

merged_table["SurveyID"] = merged_table["SurveyID"].astype(int)
merged_table["QuestionID"] = merged_table["QuestionID"].astype(int)
merged_table["UserID"] = merged_table["UserID"].astype(int)

merged_table.sort_values(by=["SurveyID", "QuestionID", "UserID"], inplace=True)
merged_table = merged_table.reset_index(drop=True)


processed_dataset = merged_table.copy()

processed_dataset.loc[
    processed_dataset["QuestionID"] == 2, "AnswerText"
] = clean_gender_data(
    processed_dataset[processed_dataset["QuestionID"] == 2], "AnswerText"
)
processed_dataset.loc[
    processed_dataset["QuestionID"] == 1, "AnswerText"
] = clean_age_data(
    processed_dataset[processed_dataset["QuestionID"] == 1], "AnswerText"
)
processed_dataset.loc[
    processed_dataset["QuestionID"] == 7, "AnswerText"
] = clean_to_numerical(
    processed_dataset[processed_dataset["QuestionID"] == 7], "AnswerText"
)
processed_dataset.loc[
    processed_dataset["QuestionID"] == 12, "AnswerText"
] = clean_to_numerical(
    processed_dataset[processed_dataset["QuestionID"] == 12], "AnswerText"
)
processed_dataset.loc[
    processed_dataset["QuestionID"].isin([28, 98]), "AnswerText"
] = clean_to_numerical(
    processed_dataset[processed_dataset["QuestionID"].isin([28, 98])], "AnswerText"
)
processed_dataset.loc[
    processed_dataset["QuestionID"].isin([56, 102]), "AnswerText"
] = clean_to_numerical(
    processed_dataset[processed_dataset["QuestionID"].isin([56, 102])], "AnswerText"
)

processed_dataset.head()

C:\Users\aga\Documents\GitHub\athiel-DA.1.5\utils\functions.py:258: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy


engine = create_engine("sqlite:///./datasets/mental_health_clean.sqlite")

processed_dataset.to_sql("Answer", engine, if_exists="replace", index=False)

engine = create_engine("sqlite:///./datasets/mental_health_clean.sqlite")


cnx2 = sqlite3.connect("./datasets/mental_health_clean.sqlite")


prism_palette = px.colors.qualitative.Prism

question_counts = (
    processed_dataset.groupby("SurveyID")["QuestionID"].nunique().reset_index()
)
user_counts = processed_dataset.groupby("SurveyID")["UserID"].nunique().reset_index()

fig_question = go.Figure(
    data=[
        go.Bar(
            x=question_counts["SurveyID"], y=question_counts["QuestionID"], width=0.2
        )
    ]
)

fig_question.update_layout(
    title="Number of Questions per Survey Year",
    xaxis_title="Survey Year",
    yaxis_title="Number of Questions",
    template="plotly_dark",
)

fig_question.show()

fig_user = go.Figure(
    data=[go.Bar(x=user_counts["SurveyID"], y=user_counts["UserID"], width=0.2)]
)

fig_user.update_layout(
    title="Number of Users per Survey Year",
    xaxis_title="Survey Year",
    yaxis_title="Number of Users",
    template="plotly_dark",
)

fig_user.show()


print("Processed Dataset")
print(processed_dataset.describe(include=["object"]).T)

Processed Dataset
               count unique                            top   freq
AnswerText    236898   4118                             -1  56110
QuestionText  236898    105              What is your age?   4218
Description   236898      5  mental health survey for 2016  88238


query_age_distribution_tech = """
    SELECT SurveyID,
        CASE 
            WHEN AnswerText BETWEEN 18 AND 24 THEN '18-24'
            WHEN AnswerText BETWEEN 25 AND 34 THEN '25-34'
            WHEN AnswerText BETWEEN 35 AND 44 THEN '35-44'
            WHEN AnswerText BETWEEN 45 AND 54 THEN '45-54'
            WHEN AnswerText BETWEEN 55 AND 64 THEN '55-64'
            WHEN AnswerText BETWEEN 65 AND 99 THEN 'Over 65 '
        END AS age_group,  
        COUNT(*) AS count
    FROM Answer
    WHERE SurveyID IN (2014, 2016, 2017, 2018, 2019)
        AND (QuestionID = 1 OR QuestionID IN (9, 13))  -- Include Question 1, 9, and 13
        AND (AnswerText > 18 OR AnswerText IN (-1, 1))  -- Include -1 and 1 for both questions
    GROUP BY SurveyID, age_group
    ORDER BY SurveyID, age_group; 
"""

age_survey_distribution_tech = pd.read_sql(query_age_distribution_tech, cnx2)

max_y_value = 800

fig = px.line(
    age_survey_distribution_tech,
    x="age_group",
    y="count",
    color="SurveyID",
    labels={"age_group": "Age Group", "count": "Count", "SurveyID": "Survey Year"},
    title="Count of Users by Age Group and Tech Employment (Questions 1, 9, and 13)",
    category_orders={
        "age_group": ["18-24", "25-34", "35-44", "45-54", "55-64", "Over 65 "]
    },
    color_discrete_sequence=prism_palette,
    template="plotly_dark",
)

fig.update_yaxes(range=[0, max_y_value])

fig.show()


query_gender_distribution = """
    SELECT SurveyID, 
        SUM(CASE WHEN AnswerText = 'Male' THEN 1 ELSE 0 END) AS male_count,
        SUM(CASE WHEN AnswerText = 'Female' THEN 1 ELSE 0 END) AS female_count,
        SUM(CASE WHEN AnswerText = 'LGBTQIA+' THEN 1 ELSE 0 END) AS lgbtqia_count,
        SUM(CASE WHEN AnswerText = 'Unknown' THEN 1 ELSE 0 END) AS unknown_count
    FROM Answer
    WHERE QuestionID == 2 OR (QuestionID == 9 AND AnswerText == 1) OR (QuestionID == 13 AND AnswerText == 1)
    GROUP BY SurveyID
"""

df = pd.read_sql(query_gender_distribution, cnx2)

prism_colors = [
    "#5f4690",
    "#1d6a96",
    "#38a6a4",
    "#0f8554",
    "#73af48",
    "#edac08",
    "#e17a05",
    "#cc513e",
    "#94346e",
    "#6f4070",
    "#666666",
]

df = df.rename(
    columns={
        "male_count": "Male",
        "female_count": "Female",
        "lgbtqia_count": "LGBTQIA+",
        "unknown_count": "Unknown",
    }
)

fig = px.bar(
    df,
    x="SurveyID",
    y=["Male", "Female", "LGBTQIA+", "Unknown"],
    labels={"value": "Count", "variable": "Gender"},
    title="Gender Distribution in Tech Surveys",
    color_discrete_sequence=prism_palette,
    template="plotly_dark",
)

fig.update_layout(
    barmode="group", xaxis_title="Survey ID", yaxis_title="Count", legend_title="Gender"
)

fig.show()


query_country_distribution = """
    SELECT SurveyID, AnswerText AS Country, COUNT(*) AS Employee_count
    FROM Answer
    WHERE QuestionID = 50
    GROUP BY SurveyID, Country
    ORDER BY SurveyID, Employee_count DESC
"""

df = pd.read_sql(query_country_distribution, cnx2)

df["Employee_count"] = df["Employee_count"].astype(int)

top_countries_by_survey = (
    df.groupby("SurveyID")
    .apply(lambda x: x.nlargest(10, "Employee_count"))
    .reset_index(drop=True)
)

fig = sp.make_subplots(
    rows=1,
    cols=1,
    subplot_titles=["Top 10 Country of Work Distribution in Tech Surveys"],
)

average_button = dict(
    label="Average",
    method="update",
    args=[{"visible": [True] * len(top_countries_by_survey)}],
)

buttons = [average_button]

for survey_id in top_countries_by_survey["SurveyID"].unique():
    button = dict(
        label=str(survey_id),
        method="update",
        args=[
            {
                "visible": [
                    survey_id == sid
                    for sid in top_countries_by_survey["SurveyID"].unique()
                ]
            }
        ],
    )
    buttons.append(button)

fig.update_layout(
    updatemenus=[
        dict(
            type="buttons",
            showactive=True,
            buttons=buttons,
            x=1,
            xanchor="right",
            y=1.15,
            yanchor="top",
        )
    ]
)

for survey_id, color in zip(
    top_countries_by_survey["SurveyID"].unique(), prism_palette
):
    data = top_countries_by_survey[top_countries_by_survey["SurveyID"] == survey_id]
    trace = go.Bar(
        x=data["Country"],
        y=data["Employee_count"],
        name=f"Survey {survey_id}",
        marker_color=color,
    )
    fig.add_trace(trace)

fig.update_layout(
    xaxis_title="Country",
    yaxis_title="Number of Employees",
    title_x=0.5,
    template="plotly_dark",
)

fig.show()


df["Employee_count"] = df["Employee_count"].astype(int)
total_respondents = df["Employee_count"].sum()

top_countries = df.groupby("Country").sum().nlargest(10, "Employee_count")

for country, count in top_countries.iterrows():
    percentage = (count["Employee_count"] / total_respondents) * 100
    print(
        f"{country}: {count['Employee_count']} respondents ({percentage:.2f}% of total)"
    )

United States of America: 1872 respondents (63.29% of total)
United Kingdom: 300 respondents (10.14% of total)
Canada: 122 respondents (4.12% of total)
Germany: 90 respondents (3.04% of total)
Netherlands: 70 respondents (2.37% of total)
Australia: 50 respondents (1.69% of total)
India: 39 respondents (1.32% of total)
France: 34 respondents (1.15% of total)
Brazil: 30 respondents (1.01% of total)
Spain: 24 respondents (0.81% of total)


query_users_participated = """
    SELECT UserID, AnswerText AS Country
    FROM Answer
    WHERE QuestionID = 50
"""

df = pd.read_sql(query_users_participated, cnx2)

total_unique_respondents = df["UserID"].nunique()

df_countries = df.groupby("Country").size().reset_index(name="Employee_count")
top_countries = df_countries.nlargest(4, "Employee_count")

for country, count in top_countries.iterrows():
    percentage = (count["Employee_count"] / total_unique_respondents) * 100
    print(
        f"{country}: {count['Employee_count']} respondents ({percentage:.2f}% of total unique respondents)"
    )

print(f"\nTotal unique respondents: {total_unique_respondents}")

73: 1872 respondents (63.29% of total unique respondents)
72: 300 respondents (10.14% of total unique respondents)
13: 122 respondents (4.12% of total unique respondents)
28: 90 respondents (3.04% of total unique respondents)

Total unique respondents: 2958


query_state_distribution = """
SELECT
    year,
    CASE WHEN state = -1 THEN 'Unknown' ELSE state END as state,
    COUNT(DISTINCT UserID) AS Employee_count,
    SurveyID
FROM (
    SELECT 
        UserID, 
        SurveyID AS year,
        MAX(CASE WHEN QuestionID == 4 AND AnswerText != -1 THEN AnswerText END) AS state,
        MAX(CASE WHEN QuestionID == 9 THEN AnswerText END) AS Tech_company,
        MAX(CASE WHEN QuestionID == 13 THEN AnswerText END) AS Tech_role,
        SurveyID
    FROM Answer
    WHERE   
        (QuestionID == 4 AND AnswerText != -1)
        OR (QuestionID == 9 AND AnswerText == 1)
        OR (QuestionID == 13 AND AnswerText == 1)
    GROUP BY UserID, SurveyID
    HAVING state IS NOT NULL AND (Tech_company IS NOT NULL OR Tech_role IS NOT NULL) 
)
GROUP BY year, state, SurveyID
ORDER BY year, Employee_count DESC;
"""

df = pd.read_sql(query_state_distribution, cnx2)

df["Employee_count"] = df["Employee_count"].astype(int)

top_states_by_survey = (
    df.groupby("SurveyID")
    .apply(lambda x: x.nlargest(10, "Employee_count"))
    .reset_index(drop=True)
)

fig = sp.make_subplots(
    rows=1,
    cols=1,
    subplot_titles=["Top 10 US States of Work Distribution in Tech Surveys"],
)

buttons = [
    dict(
        label="Average",
        method="update",
        args=[{"visible": [True] * len(top_states_by_survey)}],
    )
]

for survey_id in top_states_by_survey["SurveyID"].unique():
    button = dict(
        label=str(survey_id),
        method="update",
        args=[
            {
                "visible": [
                    survey_id == sid
                    for sid in top_states_by_survey["SurveyID"].unique()
                ]
            }
        ],
    )
    buttons.append(button)

fig.update_layout(
    updatemenus=[
        dict(
            type="buttons",
            showactive=True,
            buttons=buttons,
            x=1,
            xanchor="right",
            y=1.15,
            yanchor="top",
        )
    ]
)

for survey_id, color in zip(top_states_by_survey["SurveyID"].unique(), prism_palette):
    data = top_states_by_survey[top_states_by_survey["SurveyID"] == survey_id]
    trace = go.Bar(
        x=data["state"],
        y=data["Employee_count"],
        name=f"Survey {survey_id}",
        marker_color=color,
    )
    fig.add_trace(trace)

fig.update_layout(
    xaxis_title="State",
    yaxis_title="Number of Employees",
    title_x=0.5,
    template="plotly_dark",
)

fig.show()


df = pd.read_sql(query_state_distribution, cnx2)

df = df[df["state"].notna()]
df["Employee_count"] = df["Employee_count"].astype(int)
total_respondents = df["Employee_count"].sum()

top_states = df.groupby("state").sum().nlargest(10, "Employee_count")

for state, count in top_states.iterrows():
    percentage = (count["Employee_count"] / total_respondents) * 100
    print(
        f"{state}: {count['Employee_count']} respondents ({percentage:.2f}% of total)"
    )

California: 336 respondents (15.22% of total)
Illinois: 187 respondents (8.47% of total)
Washington: 150 respondents (6.79% of total)
New York: 123 respondents (5.57% of total)
Tennessee: 105 respondents (4.76% of total)
Texas: 98 respondents (4.44% of total)
Michigan: 96 respondents (4.35% of total)
Ohio: 95 respondents (4.30% of total)
Oregon: 84 respondents (3.80% of total)
Minnesota: 82 respondents (3.71% of total)


query_treatment = """
SELECT
    Treatment,
    SUM(CASE year WHEN 2014 THEN percentage END) AS Y2014,
    SUM(CASE year WHEN 2016 THEN percentage END) AS Y2016,
    SUM(CASE year WHEN 2017 THEN percentage END) AS Y2017,
    SUM(CASE year WHEN 2018 THEN percentage END) AS Y2018,
    SUM(CASE year WHEN 2019 THEN percentage END) AS Y2019
FROM (
    SELECT year, Treatment,
        COUNT(*) * 100.0 / SUM(COUNT(*)) OVER (PARTITION BY year) AS percentage
    FROM (
        SELECT SurveyID AS year, 
            MAX(CASE WHEN (QuestionID == 13 AND AnswerText == 1) THEN AnswerText END) AS Tech_company,
            MAX(CASE WHEN (QuestionID == 9 AND AnswerText == 1) THEN AnswerText END) AS Tech_role,
            MAX(CASE WHEN (QuestionID == 7) THEN AnswerText END) AS Treatment,
            COUNT (*) as count
        FROM Answer
        WHERE QuestionID IN (7, 9, 13)
        GROUP BY UserID, SurveyID
        HAVING (Tech_company IS NOT NULL OR Tech_role IS NOT NULL)
    ) AS t
    GROUP BY year, Treatment
) AS t2
GROUP BY Treatment
"""

df = pd.read_sql(query_treatment, cnx2)
df["Treatment"] = pd.to_numeric(df["Treatment"])

title = "Seeking Mental Health Treatment in the Past"
xaxis_title = "Treatment Response"
column = "Treatment"

create_bar_plot(df, column, title, xaxis_title)


fig = go.Figure()

query_interview = """
SELECT
    Interview,
    SUM(CASE year WHEN 2014 THEN percentage END) AS Y2014,
    SUM(CASE year WHEN 2016 THEN percentage END) AS Y2016,
    SUM(CASE year WHEN 2017 THEN percentage END) AS Y2017,
    SUM(CASE year WHEN 2018 THEN percentage END) AS Y2018,
    SUM(CASE year WHEN 2019 THEN percentage END) AS Y2019
FROM (
    SELECT year, Interview,
        COUNT(*) * 100.0 / SUM(COUNT(*)) OVER (PARTITION BY year) AS percentage
    FROM (
        SELECT SurveyID AS year, 
        MAX(CASE WHEN (QuestionID == 13 AND AnswerText == 1) THEN AnswerText END) AS Tech_company,
        MAX(CASE WHEN (QuestionID == 9 AND AnswerText == 1) THEN AnswerText END) AS Tech_role,
        MAX(CASE WHEN (QuestionID == 12) THEN AnswerText END) AS Interview,
        COUNT (*) as count
        FROM Answer
        WHERE QuestionID IN (9, 12, 13)
        GROUP BY UserID, SurveyID
        HAVING (Tech_company IS NOT NULL OR Tech_role IS NOT NULL) 
    ) AS t
    GROUP BY year, Interview
) AS t2
GROUP BY Interview
"""

df = pd.read_sql(query_interview, cnx2)
df["Interview"] = pd.to_numeric(df["Interview"])

title = "Bringing Up Mental Health Issue Over Interview"
xaxis_title = "Interview Response"
column = "Interview"

create_bar_plot(df, column, title, xaxis_title)


fig = go.Figure()

query_discussing = """
SELECT
    MAX(Discussing) AS Discussing,
    SUM(CASE year WHEN 2014 THEN percentage END) AS Y2014,
    SUM(CASE year WHEN 2016 THEN percentage END) AS Y2016,
    SUM(CASE year WHEN 2017 THEN percentage END) AS Y2017,
    SUM(CASE year WHEN 2018 THEN percentage END) AS Y2018,
    SUM(CASE year WHEN 2019 THEN percentage END) AS Y2019
FROM (
    SELECT year, MAX(Discussing) AS Discussing,
        COUNT(*) * 100.0 / SUM(COUNT(*)) OVER (PARTITION BY year) AS percentage
    FROM (
        SELECT SurveyID AS year, 
        MAX(CASE WHEN (QuestionID == 13 AND AnswerText == 1) THEN AnswerText END) AS Tech_company,
        MAX(CASE WHEN (QuestionID == 9 AND AnswerText == 1) THEN AnswerText END) AS Tech_role,
        MAX(CASE WHEN (QuestionID == 98 OR QuestionID == 28) THEN AnswerText END) AS Discussing,
        COUNT (*) as count
        FROM Answer
        WHERE QuestionID IN (9, 13, 98, 28)
        GROUP BY UserID, SurveyID
        HAVING (Tech_company IS NOT NULL OR Tech_role IS NOT NULL) 
    ) AS t
    GROUP BY year, Discussing
) AS t2
GROUP BY Discussing
"""

df = pd.read_sql(query_discussing, cnx2)
df["Discussing"] = clean_to_numerical(df, "Discussing")

df["Discussing"] = pd.to_numeric(df["Discussing"], errors="coerce")
df_grouped = df.groupby("Discussing").sum().reset_index()
df_grouped = df_grouped.dropna(subset=["Discussing"])

display(df_grouped)

title = "Fear of Negative Consequences in Mental Health Discussions"
xaxis_title = "Discussing Response"
column = "Discussing"

create_bar_plot(df_grouped, column, title, xaxis_title)


fig = go.Figure()

query_handling = """
SELECT
    Handling,
    SUM(CASE year WHEN 2014 THEN percentage END) AS Y2014,
    SUM(CASE year WHEN 2016 THEN percentage END) AS Y2016,
    SUM(CASE year WHEN 2017 THEN percentage END) AS Y2017,
    SUM(CASE year WHEN 2018 THEN percentage END) AS Y2018,
    SUM(CASE year WHEN 2019 THEN percentage END) AS Y2019
FROM (
    SELECT year, Handling,
        COUNT(*) * 100.0 / SUM(COUNT(*)) OVER (PARTITION BY year) AS percentage
    FROM (
        SELECT SurveyID AS year, 
        MAX(CASE WHEN (QuestionID == 13 AND AnswerText == 1) THEN AnswerText END) AS Tech_company,
        MAX(CASE WHEN (QuestionID == 9 AND AnswerText == 1) THEN AnswerText END) AS Tech_role,
        MAX(CASE WHEN (QuestionID == 56 OR QuestionID == 102) THEN AnswerText END) AS Handling,
        COUNT (*) as count
        FROM Answer
        WHERE QuestionID IN (9, 13, 56, 102)
        GROUP BY UserID, SurveyID
        HAVING (Tech_company IS NOT NULL OR Tech_role IS NOT NULL) 
    ) AS t
    GROUP BY year, Handling
) AS t2
GROUP BY Handling
"""

df = pd.read_sql(query_handling, cnx2)
df["Handling"] = pd.to_numeric(df["Handling"])


title = "Witnessed Poor Handling of Mental Health Issues at Work"
xaxis_title = "Handling Response"
column = "Handling"

create_bar_plot(df, column, title, xaxis_title)


filtered_db_path = "./datasets/mental_health_clean_filtered.sqlite"
valid_question_ids = [1, 2, 4, 7, 9, 12, 13, 28, 50, 56, 98, 102]

df_combined = database_filtering(
    processed_dataset, valid_question_ids, filtered_db_path
)

Columns in create_country_subquery: Index(['UserID', 'Country'], dtype='object')
Columns in create_state_subquery: Index(['UserID', 'State'], dtype='object')
Columns in create_treatment_subquery: Index(['UserID', 'Treatment'], dtype='object')
Columns in create_interview_subquery: Index(['UserID', 'Interview'], dtype='object')


print(f"{df_combined['Age'].unique()}\n")
print(f"{df_combined['Gender'].unique()}\n")
print(f"{df_combined['Treatment'].unique()}\n")
print(f"{df_combined['Interview'].unique()}\n")
print(f"{df_combined['Discussing'].unique()}\n")
print(f"{df_combined['Handling'].unique()}\n")

[37 32 31 33 35 39 42 23 29 36 27 46 34 30 40 38 50 24 18 28 26 22 19 25
 44 45 43 41 21 56 60 54 55 48 20 57 58 47 62 51 49 53 61 52 63 66 59 65
 70 64 67]

['Female' 'Male' 'LGBTQIA+' 'Unknown']

[1 0]

[0.0 1.0 0.5]

[0.0 1.0 0.5 -1.0]

[0.0 1.0 0.5 -1.0]


point_biserial_corr, p_value = pointbiserialr(
    df_combined["Treatment"], df_combined["Age"]
)

print(
    f"Point-Biserial Correlation: {point_biserial_corr}, p-value: {p_value} for Age vs Treatment ('Seeking Mental Health Treatment in the Past')"
)

Point-Biserial Correlation: 0.06802000977688193, p-value: 1.8625509518119982e-05 for Age vs Treatment ('Seeking Mental Health Treatment in the Past')


spearman_corr, p_value = spearmanr(
    df_combined["Age"], df_combined["Interview"].astype(float)
)

print(
    f"Spearman Rank Correlation: {spearman_corr}, p-value: {p_value} for Age vs Interview ('Bringing Up Mental Health Issue Over Interview')"
)

Spearman Rank Correlation: 0.0013022290674804377, p-value: 0.9347585190895162 for Age vs Interview ('Bringing Up Mental Health Issue Over Interview')


spearman_corr, p_value = spearmanr(df_combined["Age"], df_combined["Discussing"])

print(
    f"Spearman Rank Correlation: {spearman_corr}, p-value: {p_value} for Age vs Discussing (Fear of Negative Consequences in Mental Health Discussions)"
)

Spearman Rank Correlation: 0.13468394108357834, p-value: 1.8137854566051697e-17 for Age vs Discussing (Fear of Negative Consequences in Mental Health Discussions)


spearman_corr, p_value = spearmanr(df_combined["Age"], df_combined["Handling"])

print(
    f"Spearman Rank Correlation: {spearman_corr}, p-value: {p_value} for Age vs Handling (Witnessed Poor Handling of Mental Health Issues at Work)"
)

Spearman Rank Correlation: 0.11947128064109927, p-value: 4.8058393227034167e-14 for Age vs Handling (Witnessed Poor Handling of Mental Health Issues at Work)


male_treatment = df_combined[df_combined["Gender"] == "Male"]["Treatment"]
female_treatment = df_combined[df_combined["Gender"] == "Female"]["Treatment"]
lgbtqia_treatment = df_combined[df_combined["Gender"] == "LGBTQIA+"]["Treatment"]
unknown_treatment = df_combined[df_combined["Gender"] == "Unknown"]["Treatment"]

f_statistic, p_value = f_oneway(
    male_treatment, female_treatment, lgbtqia_treatment, unknown_treatment
)

print(
    f"ANOVA F-statistic: {f_statistic}, p-value: {p_value} for Gender vs Treatment ('Seeking Mental Health Treatment in the Past')"
)

ANOVA F-statistic: 56.47830940124604, p-value: 9.657269285296993e-36 for Gender vs Treatment ('Seeking Mental Health Treatment in the Past')


tukey_results = pairwise_tukeyhsd(df_combined["Treatment"], df_combined["Gender"])

print(tukey_results)

  Multiple Comparison of Means - Tukey HSD, FWER=0.05   
========================================================
 group1   group2  meandiff p-adj   lower   upper  reject
--------------------------------------------------------
  Female LGBTQIA+    0.083 0.3787 -0.0503  0.2164  False
  Female     Male  -0.2173    0.0 -0.2639 -0.1707   True
  Female  Unknown  -0.0545 0.8985 -0.2555  0.1465  False
LGBTQIA+     Male  -0.3003    0.0 -0.4295 -0.1711   True
LGBTQIA+  Unknown  -0.1375 0.4328 -0.3719  0.0969  False
    Male  Unknown   0.1628   0.15 -0.0355  0.3611  False
--------------------------------------------------------


df_combined["Gender_Code"] = df_combined["Gender"].map(
    {"Male": 1, "Unknown": 2, "Female": 3, "LGBTQIA+": 4}
)
mean_values = df_combined.groupby("Gender_Code")["Treatment"].mean()

x_values = df_combined["Gender_Code"]
y_values = df_combined["Treatment"]

m, b = np.polyfit(x_values, y_values, deg=1)
y_pred = m * x_values + b
r_squared = r2_score(y_values, y_pred)

fig = go.Figure()

fig.add_trace(
    go.Scatter(
        x=df_combined["Gender_Code"],
        y=df_combined["Treatment"],
        mode="markers",
        name="Data Points",
        text=df_combined["Gender_Code"],
    )
)

fig.add_trace(
    go.Scatter(
        x=df_combined["Gender_Code"],
        y=m * df_combined["Gender_Code"] + b,
        mode="lines",
        name="Regression Line",
        text=[f"Y = {m:.2f}X + {b:.2f}" for _ in df_combined["Gender_Code"]],
    )
)

fig.update_layout(
    hovermode="closest",
    hoverlabel=dict(bgcolor="white", bordercolor="black"),
    xaxis_title="Gender Code",
    yaxis_title="Treatment",
    title="Scatter Plot with Regression Line",
    template="plotly_dark",
)

fig.show()


contingency_table = pd.crosstab(df_combined["Interview"], df_combined["Gender"])
chi2, p, _, _ = chi2_contingency(contingency_table)

n = df_combined.shape[0]

categories_gender = len(df_combined["Gender"].unique())
categories_interview = len(df_combined["Interview"].unique())

cramers_v = np.sqrt(chi2 / (n * (min(categories_gender, categories_interview) - 1)))

print(
    f"Chi-square value: {chi2}, p-value: {p} for Gender vs Interview ('Bringing Up Mental Health Issue Over Interview')"
)
print(f"Cramér's V: {cramers_v}")

Chi-square value: 89.0708561088282, p-value: 4.7248453716916386e-17 for Gender vs Interview ('Bringing Up Mental Health Issue Over Interview')
Cramér's V: 0.10612910040159637


observed = pd.crosstab(df_combined["Gender"], df_combined["Discussing"])
chi2, p, _, _ = chi2_contingency(observed)

n = observed.sum().sum()
phi2 = chi2 / n
cramers_v = sqrt(phi2 / min(observed.shape[0] - 1, observed.shape[1] - 1))

print(
    f"Chi-square value: {chi2}, p-value: {p} for Gender vs Discussing (Fear of Negative Consequences in Mental Health Discussions)"
)
print(f"Cramér's V: {cramers_v}")

Chi-square value: 61.436824213018426, p-value: 7.080333712035073e-10 for Gender vs Discussing (Fear of Negative Consequences in Mental Health Discussions)
Cramér's V: 0.07196733474074374


observed = pd.crosstab(df_combined["Gender"], df_combined["Handling"])
chi2, p, _, _ = chi2_contingency(observed)

n = observed.sum().sum()
phi2 = chi2 / n
cramers_v = sqrt(phi2 / min(observed.shape[0] - 1, observed.shape[1] - 1))

print(
    f"Chi-square value: {chi2}, p-value: {p} for Gender vs Handling (Witnessed Poor Handling of Mental Health Issues at Work)"
)
print(f"Cramér's V: {cramers_v}")

Chi-square value: 100.70542639162835, p-value: 1.1328077252907465e-17 for Gender vs Handling (Witnessed Poor Handling of Mental Health Issues at Work)
Cramér's V: 0.09213984481481263


cnx.close()
cnx2.close()

	questiontext	questionid
0	What is your age?	1
1	What is your gender?	2
2	What country do you live in?	3
3	If you live in the United States, which state ...	4
4	Are you self-employed?	5

	Discussing	Y2014	Y2016	Y2017	Y2018	Y2019
0	-1.0	0.000000	17.860301	17.610063	20.448179	21.694915
1	0.0	41.707081	6.896552	6.446541	6.442577	4.406780
2	0.5	36.469447	45.092838	44.025157	43.137255	41.016949
3	1.0	21.823472	30.150309	31.918239	29.971989	32.881356

¶

¶

¶

¶

¶

¶

¶

Chosen Strategy for Organizing Tables:¶

¶

¶

¶

¶

¶

¶

¶

¶

¶

¶

¶

¶

	SurveyID	Description
0	2014	mental health survey for 2014
1	2016	mental health survey for 2016
2	2017	mental health survey for 2017
3	2018	mental health survey for 2018
4	2019	mental health survey for 2019